Peak calling a general bed file (not reads)

Question

[jrose]

Hi,

I have a bed file of annotated regions. I want to find genomic regions which are particularly dense with these annotations. I thought perhaps a way to do this is standard peak calling (Poisson model) and I was wondering if there are any tools out there that do this? Using MACS2 with the input bed file does not work because it keeps predicting that the read size (annotation feature size in my case) is 200. Any suggestions?

Answer 1

[jared.andrews07]

Seems like potential use-case for bedtools merge assuming these regions overlap. Merge and count regions.

Otherwise, a binning approach makes sense. Break genome up into 10 kb chunks, or whatever chunk size makes sense, intersect your regions with them, spit out counts for each bin, grab the top X of them or set an arbitrary threshold, etc.